Cookies
O website necessita de alguns cookies e outros recursos semelhantes para funcionar. Caso o permita, o INESC TEC irá utilizar cookies para recolher dados sobre as suas visitas, contribuindo, assim, para estatísticas agregadas que permitem melhorar o nosso serviço. Ver mais
Aceitar Rejeitar
  • Menu
Publicações

Publicações por Gonçalo Leão

2019

Using simulation games for traffic model calibration

Autores
Leão, G; Ferreira, J; Amaro, P; Rossetti, RJF;

Publicação
17th International Industrial Simulation Conference 2019, ISC 2019

Abstract
Microscopic simulation requires accurate car-following models so that they can properly emulate real-world traffic. In order to define these models, calibration procedures can be used. The main problem with reliable calibration methods is their high cost, either in terms of the time they need to produce a model or due to high resource requirements. In this paper, we examine a method based on virtual driving simulation to calibrate the Krauß car-following model by coupling the Unity 3D game engine with SUMO. In addition, we present a means based on the fundamental diagrams of traffic flow for validating the instances of the model obtained from the calibration. The results show that our method is capable of producing instances with parameters close to those found in the literature. We conclude that this method is a promising, cost-efficient calibration technique for the Krauß model. Further investigation will be required to define a more general approach to calibrate a broader range of car-following models and to improve their accuracy. © 2019 EUROSIS-ETI.

2023

Using Deep Reinforcement Learning for Navigation in Simulated Hallways

Autores
Leao, G; Almeida, F; Trigo, E; Ferreira, H; Sousa, A; Reis, LP;

Publicação
2023 IEEE INTERNATIONAL CONFERENCE ON AUTONOMOUS ROBOT SYSTEMS AND COMPETITIONS, ICARSC

Abstract
Reinforcement Learning (RL) is a well-suited paradigm to train robots since it does not require any previous information or database to train an agent. This paper explores using Deep Reinforcement Learning (DRL) to train a robot to navigate in maps containing different sorts of obstacles and which emulate hallways. Training and testing were performed using the Flatland 2D simulator and a Deep Q-Network (DQN) provided by OpenAI gym. Different sets of maps were used for training and testing. The experiments illustrate how well the robot is able to navigate in maps distinct from the ones used for training by learning new behaviours (namely following walls) and highlight the key challenges when solving this task using DRL, including the appropriate definition of the state space and reward function, as well as of the stopping criteria during training.

  • 2
  • 2